NYC Compression

How to do this, that and the other. Share, learn, teach. How did X do that? How can I sound like Y?
Post Reply New Topic
RELATED
PRODUCTS

Post

I probably don't use this as much as I should as I'm a little confused as to whether I should place an eq before or after the compressor on the parallel channel.

What do you do? And for what reasons?

I used to eq after the comp as the extreme compression settings made the sound boxy. So then I would boost the lows and highs to bring them back up.

But now am I right in thinking that if I eq before the comp then this will achieve frequency enhancement? So for example if I boost the highs then this will enhance the highs?
Cubase user, House producer.

http://soundcloud.com/gavin-jackson

Post

My thinking would be both pre- and post-EQ compressor.

Depending on compressor, the threshold is triggered on different frequency range.
So pre-eq makes the compressor work differently.

Then you might want to correct EQ post compressor.

Anything goes that sound good..

Post

I haven't used this method, but read about it in The Mixing Engineer's Handbook.

His recommendation is send drums and bass to another stereo channel, compress that track a lot (he mentions 10 db) followed by an EQ where you add some high end and low end. (which is optional) Then, bring the compressed channel level up under the original channel's until you can "just hear it"

Post

GFunk wrote:I used to eq after the comp as the extreme compression settings made the sound boxy. So then I would boost the lows and highs to bring them back up.
"Boxy", huh? That sounds sounds like undesirable side effect of recombining the two waves -- can you spell "comb filter"? Ooops, I just gave you the answer.

https://www.gearslutz.com/board/music-c ... n-daw.html

So by itself, unless a compressor is also doing non-linear frequency stuff... which I guess can happen with harmonic coloration and the like... the compressed signal should phase align with the original signal.

(Actually, I need to look into this myself -- is compression by itself inherently linear-phase? Hmm... I think it should be, but most useful compressors today I think do their own non-linear mojo/voodoo processing. Anyway...)

Problem is that if, for whatever reason, the original and processed signals are not phase aligned, recombining them will result in phase issues... particularly comb filtering where frequencies cancel themselves out. Ok, so that would explain how you might get a boxy sound.

So what can cause the phase alignment issues:

1.) The plugin itself is introducing phase shifts due to non-linear mojo/voodoo operations

2.) Your DAW's latency compensation is broken

3.) Your DAW's latency compensation is NOT broken, but the compressor is not reporting it's latency correctly (this happens more than we'd like, actually)

4.) Some combination of 2 & 3

Also, and FYI, just throwing EQ on top is simply masking an underlying problem. That's OK, I suppose, but I would rather fix the problem at the source. :tu:

How could we fix a phase problem? First, let's try to see if we have one...

* Isolate the source track and the parallel buss to it's own project. Or just mute everything else.

* Actually, before proceeding -- verify that the source track and the parallel buss cancel!!! Disable all plugs on the parallel bus, flip the phase of the buss -- you should hear absolutely zero audio on playback: the source and parallel bus should absolutely, 100% cancel each other due to flipped phase of the buss. If not, you know where your problem is. BTW, IIRC older versions of Pro Tools introduced latency on busses.

* OK, source track and parallel buss are aligned... now do the same phase flip test, but this time enable the compressor on the buss. Start by dialing back the compression settings at first, you should hear very close to no output due to phase flip (or maybe just a little faint something). Now start dialing up the compression settings, particularly any mojo/voodoo inducing parameters. Do you hear lots of output... like more than just a faint difference? If so, all of that difference you're hearing is what's being subtracted OUT of your mixed signal when both source and buss are set to the same phase.

* If you hear significant phase problems with your compressor, try the same test except with a compressor known to be "transparent"... like Tonebooster's Compressor, or any number of others known to be transparent. It's important that it's transparent.

Possible workarounds:

1.) Use a compressor that is handled properly wrt. latency compensation. Any unaccounted signal delay will most definitely cause phase alignment issues. This is the first thing to check and rule out/fix!

2.) Run the same phased flip test above, except on the dry signal (after being sent to the buss) add an instance of the same compressor you use on the buss... set the compression on the dry track to 0%. The idea here is to pass the dry signal through the same "circuitry" of the compressed buss, just without any actual compression. Note: DO NOT "bypass" the plugin, just effectively disable compression on the source track. Basically, any phase issues introduced by the plugin itself will be introduced in both source and parallel buss, so the end result is... er... no phase issues. I this case, two wrongs most definitely make one right! Does that make sense?

3.) Use a compression plugin that has it's own mix knob... There are a handful these days that'll do the parallel compression for you ITP (In The Plug), as it were. For example, FabFilter Pro-C, The Glue, etc.
You need to limit that rez, bro.

Post

Kbaccki, that is a great explanation. Very well done.

I only wish to add-- All dynamics processors are classed as nonlinear devices.

A linear device always does exactly the same thing to a signal regardless whether the signal is big or little, bassy or bright. An input at -20 dB would have the exact same output, as the exact same input at 0 dB. The only difference the linear device would contribute, is that the loud input signal would have an output 20 dB louder. The two output signals would be identical except that the loud input would have 20 dB hotter output than the quiet input.

No dynamics processor that I know about, can fit this definition of a linear process. The very purpose is to dynamically change the output in some way, depending on changes of the input.

I'm being pedantic in a likely failed attempt to include such as de-essers, dynamic EQ's, enhancers, noise reduction units, envelope tracking filters. All of which are non-linear even if carefully designed to sound as clean and transparent as possible, and might sometimes linearly preserve the input-to-output gain for all signals.

They are all nonlinear because they adapt what they do, according to the nature of the input.

Linear devices always do exactly the same thing to the input signal regardless of changed properties of the input. Example linear devices might include straightforward "idealized" EQ, gain-scaling, echo, delay, reverb, mixing, samplerate conversion, etc.

So the trick if one wishes to design a clean-sounding compressor, is to design the nonlinear device so that the ear doesn't perceive the nonlinear behavior as distortion.

Post

Thank you kbaccki for your detailed response. I'm afraid it is not a comb filtering problem, my DAW is operating fine. What I was referring to when I said "boxy" was when you compress something hard the louder frequencies get attenuated more (these being the bass frequencies in particular) and the sound becomes more "middley" or "boxy".

What I'm really trying to establish why you would place an eq before the compressor the the parallel channel?

Also come to think about it why hasn't someone come up with the idea of a plugin yet where you can do parallel compression and also eq the parallel channel? A kind of all in one plugin.

I know and love Pro-C but you can only filter the sidechain you can't eq the parallel channel.
Cubase user, House producer.

http://soundcloud.com/gavin-jackson

Post

I've only used parallel compression to beef up vocals, and it was quite awhile ago.

The last application was to get more intelligibility on a strong baritone over dense rock. Squash the compressed branch, then cut the bass and raise upper mids, and add just enough of that to the full-bodied, lightly compressed, minimal EQ main branch.

For instance in that case, if low vocal notes were lots louder than the high notes (or vice versa), unless the eq were applied pre compression, then the intelligibility branch would fade and rise according to the pitch of the note being sung. The goal being to get equal intelligibility on all notes.

But in general I'd be inclined to try post-eq first, and then try pre-eq if post-eq wasn't working out. That preference I can offer no reasonable explanation for. Maybe pre-eq would always be the no-brainer first choice to try.

Thinking about it, makes me wonder why such parallel processing doesn't almost always make it sound worse and not better. EQ virtually always plays hob with phase, which the ear tends not to notice if all you hear is a single path post-EQ.

But if you mix the EQ output with a straight copy, or a different-EQ'd copy, the relative cancellations could turn the end result into a big mess of a difficult-to-predict funky comb filtering, and the comb would mutate every time you tweak an EQ knob, and the comb would change as the compressor pumps along, dynamically changing the mix between dry and EQ.

Maybe what makes it practical at all, is that usually the compressed branch would be set to a much lower level than the main branch? So as to minimize the dynamic combing?

There are some EQ's which can modify freq balance without corresponding phase shift, but they accomplish the trick by changing the group delay of the various frequencies in a fashion different from the original signal. So there might not be any free lunch, because the frequency-dependent timing shifts could also cause comb filtering when mixed with the unprocessed track.

Post

Thanks JCJR.

Any other opinions on this?.......

What got me wondering about this was a Waves video for their V series bundle. The mixing engineer is using the V comp for parallel compression and he preceeds this with the V eq4 on the parallel channel.
Cubase user, House producer.

http://soundcloud.com/gavin-jackson

Post

GFunk wrote:What I'm really trying to establish why you would place an eq before the compressor the the parallel channel?
Because that will change compressor response. I.e. if you boost the mid-range in bass heavy sound then flatten the signal with compressor, it will apply more compression to the mid-range. While if you feed compressor original signal it will be mostly reacting to the bass frequencies, leaving mid-range less compressed.
And if you apply mid-range boost post-compression, you will boost less compressed mid-range signal, while still having more compressed bass frequencies.

You might imagine it like this:
Mid range EQ boost pre-compressor
Hi -> Compression -> Less compressed -> Less loud
Mid ->EQ Boost -> Compression -> More compressed -> Louder
Low -> Compression -> Less compressed -> Loud

Mid range EQ boost post-compressor
Hi -> Compression -> Even less compressed -> Less loud
Mid -> Compression -> Less compressed -> EQ Boost -> Louder
Low -> Compression -> More compressed -> Loud
Wonder whether my advice worth a penny? Check my music at Soundcloud and decide for yourself.
re:vibe and Loki Fuego @ Soundcloud

Post

All the EQ is doing is boosting or attenuating certain frequencies.
This will cause these frequencies to hit the compressor threshold harder or softer.
So, you can in reality change the compressor's reaction to the material.

Example:
Maybe the Kick Drum is hitting the compressor too hard and makes it over react.
Cutting some lowend, pre compressor, will allow the compressor to react more
naturally. If that is what you want.
This is probably one of the most common uses of EQ before compression.
You can take it the other way too. Boosting the low end will cause
more compression on the kick.

I would rather use a compressor with a side chain filter for this though.

Post

GFunk wrote:I probably don't use this as much as I should as I'm a little confused as to whether I should place an eq before or after the compressor on the parallel channel.

What do you do? And for what reasons?

I used to eq after the comp as the extreme compression settings made the sound boxy. So then I would boost the lows and highs to bring them back up.

But now am I right in thinking that if I eq before the comp then this will achieve frequency enhancement? So for example if I boost the highs then this will enhance the highs?
You can do it both ways....

For more accurate compression, place the EQ before the compressor.

For more "coloration", place the EQ after the compressor.

In some cases where tight dynamics are important, you can place a brick wall limiter after the compression.
ONLY allow the absolute peaks (1dB or less) to hit the limiter.

This is an effective way to get transparent compression.

Post

Lin Phase eq before camp
eq after fomp.
Depends on if you want eq to impact the comp behavior.

Post Reply

Return to “Production Techniques”